A Multi Technique Study of Fluorinated Nanodiamonds for Low-Energy Neutron Physics Applications

Data of quasi-specular reflection of cold neutrons, prompt-γ neutron analysis, X-ray Raman scattering (XRS), and neutron pair distribution function (PDF) analysis with powder of detonation nanodiamonds are analyzed to collect their structural properties and chemical composition. Both as-synthesized and purified samples were studied using fluorination samples. Removal of both the sp2 amorphous carbon shell and the hydrogen atoms is evidenced respectively by the change of neutron-nuclei optical potentials of nanoparticles and the increase of their neutron reflectivity. Moreover, sp3 diamond cores of nanoparticles stay intact during the fluorination as revealed by similar scattering patterns, PDF, and XRS data. Quasi-specular reflection, PDF, and XRS data are complementary for the study of nanomaterials and in good agreement with conventional characterization techniques (infrared spectroscopy and solid-state NMR).


Fine investigation of nanomaterials needs a combination of complementary characterization
techniques in order to collect information at different scales concerning their structure and chemical composition. The perfect knowledge of those characteristics allows a rational use of the nanomaterials. In the present paper, we apply non-conventional techniques to characterize nanodiamonds (NDs) powders 1,2 , after purification under molecular fluorine flux 3,4 . As a matter of fact, nanodiamond powder combines high volume density of diamond, high coherent scattering length (b C = 6.65 fm), low neutron absorption (σ abs C = 3.5 mb) and inelastic scattering cross-sections of C. NDs can be efficiently used in low-energy neutron physics due to the quasi-3 specular reflection of cold neutrons (CNs) at small incidence angle to powder surface 5,6,7,8 and the diffusive reflection of very cold neutrons (VCNs) at any incidence angle 9,10,11,12 .
Cross-sections of neutron scattering on NDs have been studied in detail 13,14 . These investigations compare, in particular, different techniques of producing NDs and measure temperature dependence of inelastic scattering cross-sections. The authors show virtual absence of low-energy excitations thus highly elastic neutron scattering and underline the importance of clustering or agglomeration of nanoparticles in powders for neutron diffusion and transport.
Important efforts have been devoted for including the diffusion of slow neutrons in the DND powder to neutron transport simulations 15,16 .
X-ray Raman Scattering (XRS) and Pair Distribution Function analysis (PDF) are both bulk techniques very sensitive to the atomic local structure and C hybridization. PDF is a wellestablished technique for studying disordered or semi-ordered materials while XRS is a nonresonant inelastic X-ray scattering technique that allows observing the core electron excitation in the sample and thus the unoccupied electronic density of states 17,18 . XRS uses X-rays in the hard regime, giving a penetration depth in the order from hundreds of micrometers to millimeters.
Prompt-γ neutron analysis allows the absolute evaluation of hydrogen content 19 . Neutrons scattering represents a powerful method 20,21,22,23 to study the size, shape and positions of scatter.
Quasi-specular reflection of CNs may be used for the general purpose of characterization of nanomaterials in addition to conventional techniques such as XRD, Raman and solid-state NMR.
NDs can be produced in the nanocrystalline form, by the detonation of carbon-containing explosives such as trinitrotoluene and hexogen in a steel chamber 1,24  In order to probe the local structure in the core of the NDs, we have employed XRS and PDF of neutron diffraction. PDF provides straightforward information about the local structure in terms of first neighboring distances. XRS allows the direct observation of soft X-rays edges (like C and F K edges) of bulk samples thanks to the use of hard X-rays that are inelastically scattered 17,18 .
Both techniques are therefore sensitive to the C-hybridization and local structure with a bulk penetration depth. Data from quasi-specular reflection of cold neutrons, PDF and XRS will be compared to conventional techniques such as XRD, solid state NMR, Raman and infrared spectroscopies in order to clarify in great details the structure of NDs and evidence the efficiency of the new techniques. Ab initio theoretical calculations for bulk diamond and the diamond (111) H-terminated surface were performed within the framework of the density functional theory (DFT).

2.1
Materials. NDs were fluorinated in static conditions using a two-step process in a closed Nickel reactor. NDs were placed onto passivated Nickel supports (covered with NiF 2 ) inside the reaction vessel. Prior to the fluorine gas insertion, the reactor was evacuated to primary vacuum (∼10 -2 mbar) for 12 hours. A flux of pure fluorine gas (99.9% purity) was then added to reach 0.6 atm inside the reactor. This condition was fixed in order to avoid pressure higher than 1.2 atm during heating and decomposition gas evolution resulting from the fluorination. The temperature was increased to 450 °C and then stabilized for 12 h. After this first treatment and cooling to room temperature, the reactor was flushed with nitrogen flow for 1 hour to remove unused fluorine, HF and decomposition products (CF 4 , C 2 F 6 , …) 28 . For the completion of both sp 2 layer and hydrogen removal, the same vacuum/fluorine insertion/fluorination at 450 °C for 12h/evacuation/cooling sequence was carried out in the second step. On the one hand, with such a two-step treatment, amorphous carbons located onto the ND surface must totally decompose into gaseous CF 4 and C 2 F 6 ; weight loss of the sample was then expected. On the other hand, formation of covalent C-F bonds resulted in weight uptake. Finally, the weight change is very low during the treatment, less than 2 wt%, also highlighting the stability of the NDs under these drastic conditions. The resulting samples are then denoted F-NDs.

Neutron Diffraction Data for PDF Analysis.
Neutron diffraction data (more details can be found in 29 ) were collected at the neutron diffractometer for disordered materials D4C at 6 the ILL (Institut Laue Langevin, Grenoble, FR) 30 . We used an incident wavelength λ= 0.4989 Å corresponding to a maximum Q of 23.5 Å -1 . Powder samples were placed inside a sealed 7 mm diameter Vanadium cylindrical cell and measured at room temperature in a vacuum. In order to obtain the PDF function, prior to the space-Fourier transform, an appropriated background and multiple corrections were applied to the raw data using the CORRECT program 31 . All the data reduction and treatment were done using the routines available on the instruments 32 . beam angle was 10 degrees. All data were collected at room temperature. Raw data from pixelated images were treated using the XRS tools python routines, as described elsewhere 35 .

Quasi-Specular Reflection, Neutron Prompt-Analysis and Conventional
Methods. Quasi-specular reflection of cold neutrons was measured at D17 neutron reflectometer 36 at ILL. Experimental details are given in reference 7 .
In order to measure very small residual amounts of H after the powder purification, we applied the neutron prompt-analysis method as in 37 but used a much more intense neutron beam (PF1B instrument at the ILL 38 ). To minimize the background in this measurement, we, on one hand, placed powder samples in thin-walls aluminum envelopes with a low hydrogen content, and, on the other hand, carefully protected the germanium γ-detector against scattered neutrons and γ-quanta. The count-rate in the -quanta peak of total absorption of the reaction allowed measuring content in the sample and in the empty envelope used as background. The absolute calibration of the -quanta detection efficiency in the H peak was carried out using a thin polyethylene sample with a precisely known amount of .
FTIR spectrometer NICOLET 5700 (Thermo Electron) was used to record IR spectra using both ATR and transmission modes. 100 scans with 4 cm -1 resolution were collected to acquire each spectrum between 4000 and 400 cm -1 . The single-reflection ATR accessory (Thermo Scientific Smart Orbit) is working with a durable diamond crystal (type IIa Diamond tungsten carbide mounted in stainless steel with a refractive index of 2.4 and an incident angle of 58°) and a swivel pressure tower that ensured consistent pressure from sample to sample. The active sample area was 1.5 mm 2 . A wide spectral range (10000 to below 200 cm -1 ) and good depth of penetration (DP of 2.03 µm at 1000 cm -1 ) were then reached. NMR experiments were carried out on a Bruker Avance spectrometer with operating frequency of 282.2 19 F. A simple sequence (τ-acquisition) was used with a single π/2 pulse length of 5.5 μs and the recycle time was equal to 5 s. 19 F chemical shifts were referenced to CFCl 3 . We have also carried out calculations of the core level binding energies for the C-1s core state, including final-state core hole effects (core-level relaxation energy), which involve the recalculation of the Kohn-Sham eigenvalues for the core states for the chosen atom using a half core-hole 43

RESULTS AND DISCUSSION
Prompt-analysis showed that content in NDs is drastically reduced by the fluorination, achieving the level of , which is 35-60 times lower than that before fluorination. The H/C atomic ratio are equal to and for NDs and F-NDs respectively. Quantitative NMR using polytetrafluoroethylene (PTFE) as an internal reference (Fig. 1)

Structural Characterization.
Neutron PDF data reveals the maintaining of a diamond core after fluorination (Fig. 2) in good accordance with X-ray diffraction 3

X-ray Raman Scattering and Spectroscopic Characterization. XRS was used
to investigate the electronic and local atomic structure of NDs, before and after the fluorination process. XRS spectra at the C K-edge and F K-edge are presented in Figure 3.

11
The XRS spectra of graphite and diamond powders are reported from reference 29 as an example of pure Csp 2 and Csp 3 hybridization 47 . As discussed in refs 29 and 47 , the main features of the C absorption edge are associated to the transitions of the C 1s electron of the empty π and σ states in the conduction band. These states are indicated in Figure 3 as π* (in graphite only) and σ*. spectra give additional information about the total removal of OH groups (Fig. 4); the assignment of the vibration modes is reported on the figure. The broad bands in the 3280-3675 cm -1 (Fig. 4b) range and at 1640 cm -1 (Fig. 4a) Because C-H is converted into C-F during the treatment, the line is not present after the treatment. Surprisingly, the intensity of C=O band increases after fluorination. An IR spectrum of the raw sample is ill-defined due to the conductive carbonaceous shell on the diamond core and the C=O vibration band may be hindered. Nevertheless, FTIR and Raman spectroscopies evidence qualitatively the removal of C-OH and C-H groups as well as the sp 2 carbons layer onto the diamond core.

Quasi-Specular Reflection.
Here, we analyze the method of neutron quasi-specular reflection from the point of view of information about the studied samples that it can provide. 16 The main factor determining the patterns of quasi-specular reflection is neutron scattering on diamond cores of nanoparticles. Therefore, the general similarity of the patterns measured with the two studies samples (Fig. 5a and Fig. 5b) means that the fluorination effect on sp 3 cores is small or absent, in agreement with XRS.
Nevertheless, some excess in the probability of quasi-specular reflection from a fluorinated powder (Fig. 5b) over the probability of reflection from a raw powder (Fig. 5a) indicates a change in its properties. This change becomes better visible after integration the scattering patterns over the whole range of scattering angles Fig. 5c; the results for F-NDs and NDs were compared for the incidence angle of 1 o , 2 o and 3 o , respectively. Presentation of experimental data in the format of Fig. 5c is also useful for analyzing the effect of neutron diffraction in diamond cores of nanoparticles.
Diffraction diffuses neutrons to large angles thus eliminating them from quasi-specular directions. The cut-off wavelength for diamond is twice the (111) interplanar distance, or 4.12 Å.
The cut-off is not sharp due to broadening associated with finite nanoparticle sizes; as a result, the diffraction suppression extends towards ~5Å. The scale of this broadening is inversely proportional to the size of the nanoparticles. Using this simple consideration, it is straightforward to estimate a mean size of diamond sp 3 cores. However, an accurate calculation of the mean size, as well as the size distribution of diamond cores, will require the development of a theory that takes into account neutron diffraction by small size diamond nuclei of arbitrary shape, as well as the exact contribution of this effect to the probability of quasi-specular reflection of neutrons. Fig. 5c also shows that the mean difference between the scattering curves is a significantly larger probability of quasi-specular reflection of neutrons from fluorinated nanoparticle powder at the neutron wavelengths of 4-5 Å. This difference is due to the scattering of neutrons to large angles in reflection from the raw powder, in particular in the sp 2 C shells with respectively larger interatomic distances than those in the diamond. Thus, we conclude about the removal of nanoparticle shells consisting of sp 2 carbon due to fluorination.
The overall efficiency of quasi-specular reflection is improved with F-NDs. Such effect is due to fluorination, which had removed sp 2 shell from the nanoparticles and hydrogen contaminations 17 . This purification occurs without major changes in the diamond core.

CONCLUSION
A set of complementary non-conventional techniques were applied to characterize nanodiamond powders. In particular, the effect of purification using fluorination was investigated.
Fluorination process appears effective in drastically reducing the H content and sp 2 carbon layer at the surface of NDs, thought preserving the sp 3 diamond core. PDF and XRS highlight the maintaining of the diamond core during fluorination without structural change. Quasi-specular reflection of CNs from powder of raw NDs and purified NDs indicates the removal of sp 2 C layer. Prompt-γ neutron analysis evidences the drastically reduced H content. On one hand, purified NDs are confirmed to be much more efficient neutron reflectors for CNs and VCNs. On the other hand, we have shown that XRS and neutron quasi-specular reflection are powerful characterization techniques. The methods can be applied to powders of any nanoparticles (not only nanodiamonds) provided the particle size ranges from a few to a few tens of nanometers.